[teampractices] ANYONE: Phab, burnup charts, and double-counting

Thu Apr 30 00:41:33 UTC 2015

(cc'ing Christopher Johnson who wrote the Sprint/burndown extension)

On Wed, Apr 29, 2015 at 4:36 PM, Matthew Flaschen <mflaschen at wikimedia.org>
wrote:

> On 04/29/2015 11:26 AM, Kevin Smith wrote:
>
>> If an epic (or any other task) gets estimated, and then gets split into
>> subtasks which get estimated, we risk double-counting.
>>
>
It seems a general problem with any task with subtasks.  If you add a
subtask with points, then you need to reduce the parent's story points by
that amount, and since there's no history and add a comment "I added
subtask {T*NNN*} so reduced story points from 13 to 10". This is quite
prone to error. But the math can't be automated because AIUI Phabricator
doesn't distinguish subtasks from blocking tasks.

Maybe the sprint extension could add a dedicated Action > Change Story
Points to speed this up, I created T97616
<https://phabricator.wikimedia.org/T97616>.

> ...
>> I only started thinking about this an hour ago, but so far I am
>> perplexed as to how to work around this. For what it's worth, solutions
>> could include technical features, or socially-enforced conventions.
>>
>
> The latter is probably best (or at least fastest).  The simplest is just
> to set the point value to 0/empty if there are subtasks/blocking tasks.
> Instead, estimate those smaller tasks.
>

That doesn't work if you "chip away" at a piece of a task with a subtask.
If you completely break out the parent into subtasks, then it's a
degenerate case of my approach: "I fully broke this out into subtasks, so I
reduced story points to 0 from 10."

--
=S Page  WMF Tech writer
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.wikimedia.org/pipermail/teampractices/attachments/20150429/e3f36bcb/attachment.html>